Politicizing Europe in times of crisis

ABSTRACT This paper starts from the premise that the politicization of Europe is indicative of a new structuring conflict that involves a set of processes which put the national political community under strain. This structuring conflict has been emerging long before the Euro and refugee crises. However, these crises may have reinforced and potentially reshaped public conflicts within and across countries. Therefore, the paper traces the politicization of Europe during national election campaigns in fifteen countries from the early 2000s up to 2017. The analysis focuses on the way the multiple crises have affected the level of politicization, its driving forces, and the location of European issues in the political space. Overall, the results indicate a substantive increase in politicization, but they also point to strong region and crisis-specific varieties which should be considered in scholarly discussions on the relative impact of domestic conflicts on the future course of European integration.

'universalism-communitarianism' (Bornschier 2010), 'cosmopolitanism-communitarianism' (Zürn 2019), 'cosmopolitanism-parochialism' (de Vries 2018) to the 'transnational cleavage' (Hooghe and Marks 2018). However, what they all emphasize is that the new divide constitutes a break with the period of 'permissive consensus' and that conflicts over Europe have been transferred from the backrooms of political decision-making to the public sphere.
Following Hooghe and Mark's (2009) path-breaking contribution, the scholarly literature has revived the concept of politicization to describe the process of more publicly visible contestation related to the various dimensions of European integration (e.g. de Wilde et al. 2016;Hoeglinger 2016;Hurrelmann et al. 2015;Hutter and Grande 2014;Rauh 2016;Statham and Trenz 2013). As the large-scale analysis of Hutter et al. (2016) indicates, the politicization of Europe has been characterized by 'a patchwork of politicizing moments' rather than a uniform trend towards ever more politicization. It intensified during predictable institutional and policy-related events at the European level (such as Treaty changes or European summits) and the national level (such as national referendums on European issues), and it involved two types of conflicts that fed into each otherintergovernmental conflicts at the European and inter-party conflicts at the national level. Also, politicization of European integration does not seem to have been a direct result of the 'authority transfer' as suggested by de Wilde and Zürn (2012), nor is it a post-Maastricht phenomenon as some have maintained (de Wilde and Zürn 2012;Hooghe and Marks 2009). Instead, it has been flaring up and temporarily reaching impressive levels at critical moments before and after Maastricht depending on political actors and their mobilization strategies.
Crises like the Eurocrisis and the so-called refugee crisis also constitute such critical moments in the integration process that contribute to its politicization and that may reinforce the new structuring conflict (on the Eurocrisis, see Statham and Trenz 2015). As noted by van Middelaar (2016), crises are 'moments of truth', and in crises we are experiencing a 'return of politics' (as compared to the normal predominance of rule-based decisions in European politics). It is important to add that both crises concern core state powers, i.e. an integration field which raises salient issues of statehood, national community and democratic self-determination and which is, therefore, prone to be politicized and to pit national publics against one another (Genschel and Jachtenfuchs 2018: 182).
Shifting from the process of politicization to its outcomes, recent debates among integration theorists focus on how Europe's elites have dealt with the challenges posed by the two crises in a comparative perspective (Biermann et al. 2019;Börzel and Risse 2018;Genschel and Jachtenfuchs 2018;Schimmelfennig 2018). Scholars agree on the fundamental status of the crises but disagree on their outcomes and the drivers of (non-)integration (including politicization). While Genschel and Jachtenfuchs (2018) emphasize similarities in outcomes, the other cited contributions put the accent on differences: further integration in the Eurocrisis on the side and deadlock and noncompliance in the refugee crisis on the other. In turn, they emphasize the explanatory power of member state preferences (the intergovernmental answer by Biermann et al. 2019), transnational interdependence and supranational capacity (the neo-functionalist answer by Schimmelfennig 2018), varying patterns of identity politics (the refined post-functionalist answer by Börzel and Risse 2018), or the field of integration (the core state power answer by Genschel and Jachtenfuchs 2018).
For our argument, it is important that all contributions take for granted high or even 'unprecedented' levels of politicization in national arenas. Similarities in politicization in the two crises are then interpreted as indicating that domestic conflicts cannot account for differences in outcomes (Schimmelfennig 2018), as reinforcing member state preferences at best (Biermann et al. 2019) or as indicating similarities in the initial problem constellation (Genschel and Jachtenfuchs 2018). Only Börzel and Risse (2018) present more detailed arguments about the politicization process: probably higher in the refugee crisis, varying across regions, and shifting from conflicts over order in the Eurocrisis to conflicts over boundaries in the refugee crisis. However, like the other contributions, they base their claims on anecdotal evidence and previous studies, which have a limited empirical scope and, thus, do not allow to fully grasp the impact of the recent crises on politicization. Most published studies focus on countries from Northwestern Europe (where the structuring capacity of the new conflicts seem most pronounced) and they do not present long enough timelines (in particular, they do not cover the latest developments since the refugee crisis).
Given limits in empirical research and the strong claims made in EU studies about the level and (non)-effects of politicization, we take a step back in this contribution and map how the process of politicization has developed in the national electoral arena of fifteen European countries from the early 2000s to 2017. Our study includes crucial moments in both the Euro and the refugee crisis and the country selection allows assessing differences across Europe's macro regions, i.e. Northwestern, Southern, and Central-Eastern Europe. The last point is particularly important for a perspective that considers conflicts over Europe to be embedded in broader changes of national conflict structures.
Methodologically, we rely on a relational content analysis of newspaper coverage of party competition during national election campaigns. We present three pieces of evidence about (a) the systemic level of politicization, (b) the partisan divides over European integration, and (c) the configuration of the party systems in the three regions before and during the crisis showing how European issues are embedded into the political space. Note that the ambition of our study is above all descriptive, as we map the politicization of Europe in national electoral politics across contexts and over time. Most importantly, our study is limited to the extent that we do not systematically link politicization in national politics to the outcome of the two crises. In other words, we do not aim to settle the debate between the various theoretical approaches in EU studies sketched above but provide further empirical input to it. Importantly, we show that integration theorists need to incorporate not only questions of 'more or less' politicization, but also strong regional and crisis-related varieties in politicization when assessing the relative impact of domestic conflicts on the future course of European integration.

Theoretical framework
Politicization and its driving forces The literature has developed a broadly shared understanding of the concept of politicization. Zürn (2019) suggests that it can be generally defined as 'moving something into the realm of public choice', while Grande (2014: 1003) define politicization 'as an expansion of the scope of conflict within the political system'. In operational terms, a consensus is emerging regarding the components of what we mean by the term 'politicization' (de Wilde et al. 2016;Hoeglinger 2016;Hutter and Grande 2014;Rauh 2016;Statham and Trenz 2013). Accordingly, we should distinguish between three conceptual dimensions which jointly operationalize the term: issue salience (visibility), actor expansion (range) and actor polarization (intensity and direction). Note that we adhere to a definition that privileges public discourse and the supply side of politics (but cf. de Vries 2018; Hurrelmann et al. 2015). Thus we conceptually distinguish politicization from related dynamics in public opinion and individual political behavior.
For the politicization of new conflicts in general, established parties make for unlikely candidates. Drawing on cleavage theory, Hooghe and Marks (2018: 112) have aptly stated that the flexibility of established parties on major conflict dimensions is 'constrained to the extent that they have durable constituencies of voters, a decentralized decision-making structure, a self-selected cadre of activists, a self-replicating leadership, and a distinct programmatic reputation'. Other authors (e.g. Green-Pedersen 2012) have pointed out that mainstream parties have additional, strategic reasons to avoid the politicization of European integration: from internal divisions (e.g. de Vries and van de Wardt 2011; Steenbergen and Scott 2004) to responsibility once in government (e.g. Sitter 2001).
In line with this argument, the mainstream parties which have typically been pro-European have traditionally sought to depoliticize European integration in domestic arenas in many ways. Their repertory of depoliticization strategies has been vast (e.g. de Wilde and Zürn 2012; Schimmelfennig 2019). They have not only de-emphasized the issue in national elections; they also sidestepped treaty changes to avoid referendums, delegated authority to so called 'non-majoritarian' technocratic institutions, opted for Eurocompatible government formation, adopted incomplete contracts, and, most generally, resorted to integration by regulation.
These avoidance strategies of mainstream parties led Hooghe and Marks (2018) to expect that the sources of dynamism in party systems in response to major shifts in voter preferences are new parties. As new fundamental conflicts develop in society and mainstream parties try to avoid them, voters are expected to turn to new parties with distinctive profiles for their articulation. This also applies to European integration. Eurosceptic challenger parties have turned out to be the main drivers of politicization. As conflicts over Europe cut across the traditional conflict between left and right in European party systems, the Eurosceptic challenger parties have come from both ends of the political spectrum. Thus, the relationship between the traditional and the European conflict is best described as the well-known inverted Ucurve (e.g. Hooghe et al. 2002).
However, mainstream parties may ever less succeed or even want to depoliticize European integration. As Dolezal and Hellström (2016) have suggested, there is an alternative path to EU politicization, which involves the strong polarization between (mainstream) parties in government and opposition. As they show in their analysis of six Northwestern European countries, in the UK and in Germany, the politicization of Europe was driven by the competition between the government and the opposition. Similarly, Hellström and Blomgren (2016) have argued that not all mainstream parties that are internally divided on EU integration succeeded in ignoring it. As a matter of fact, they find that parties such as the British Conservatives, with a strongly divided membership and electorate, are most present in the public debate over the issue.
The jury is thus still out on which parties are driving the politicization process in the public sphere. The contributions to de Wilde et al. (2016: 11) provide little evidence for the 'strategic competition hypothesis', as they call the politicization by mainstream parties, while Hobolt and de Vries (2015) provide evidence for both strategic factors and the 'new cleavage hypothesis'. They find that 'parties that lack office-holding experience, hold non-mainstream policy positions, and have suffered electoral defeat are most likely to act as issue entrepreneurs' (p. 1177). At the same time, political losers (whether mainstream or challenger parties) are generally more likely to mobilize a previously ignored issue (such as European integration) if they are not internally divided.

Regional varieties of politicization in times of crises
It is important to keep in mind that we are (a) looking at the domestic trajectory, i.e. at politicization by inter-party competition, and (b) we consider conflicts over Europe as being embedded in the broader long-term restructuring of conflict structures. Thus, the impact of the crises on politicization is not only conditioned by the depth of the two crises and the measures taken to cope with them but also by the degree of institutionalization and structuration of the national party systems at the onset of the crises (for a more detailed argumentation, see Hutter and Kriesi 2019).
While we acknowledge that national conflict structures differ from one country to the other, we suggest that it makes sense to reduce the complexity by insisting on the respective differences between three large European regions -Northwestern Europe (NWE), Southern Europe (SE), and Centraland Eastern Europe (CEE). The countries in the three regions are not only distinct with regard to the development of their party systems, as will be become clear in the subsequent discussion, but their experience during the most recent crisis period has also been quite distinct: the Southern European countries faced both a deep economic and political crisis, while the other two regions recovered fairly quickly from the immediate economic shock of the Great Recession and most CEE countries experienced a political crisis already before the onset of the Great Recession. In addition, as Biermann et al. (2019: 256) show, countries in NWE (plus Greece and Italy) were generally more affected by increasing numbers of asylum-seekers during the refugee crisis than the 'non-affected' or transit countries in CEE.
The new structuring conflict that we described in the introduction has had the greatest impact on party politics in Northwestern Europe (NWE). There is cumulative evidence that the rise of this conflict has led to a reinterpretation of the cultural dimension of the traditionally two-dimensional space in this part of Europe (e.g. Bornschier 2010; de Vries 2018; Hooghe and Marks 2018; Kriesi et al. 2008). Ultimately, this has shaped a three-polar party configuration, pitting a unified political left against the populist radical right with the moderate right positioned in-between the two poles. In NWE, the populist radical right has been the driving force of the reconfiguration of the cultural axis. The transformation of party competition in NWE has been a long-term process that dates at least as far back as the early 1980s. The impact of the Great Recession has been short-lived in this part of Europe, and is unlikely to have fundamentally modified this long-term process. If anything, the intergovernmental conflicts between 'debtor' and 'creditor' countries that resulted from the management of the Eurocrisis are likely to have enhanced the long-term process by contributing to an increase in the politicization of Europe. The refugee crisis, by contrast, has 'hit' the countries in NWE more lastingly and is likely to have enhanced the long-term trends towards a politicization of the new structural conflicts to an even greater extent than the Eurocrisis. It served to link the twin issues of European integration and immigration in a particularly explosive way that is likely to have politicized both of them.
The new structuring conflict and its impact on the national party systems have been much weaker in SE in the pre-crisis period. By contrast, SE countries have been hit much more heavily by the Great Recession and the management of the Eurocrisis had a much stronger impact on their electoral politics. These countries were ill-prepared to deal with the crisis. Structural problems (weak state capacity resulting from widespread clientelistic practices and corruption), policy errors and misconceptions predated the Eurocrisis, left them vulnerable to the economic shock and led to the intervention of European actors These actors (the 'Troika') intervened directly in domestic economic policy-making and (in Greece) even in national politics. The pressure was strongest on Greece and Portugal, but even Italy and Spain felt the heat from European actors and became the object of 'implicit' if not of formal conditionality (Sacchi 2014). Ultimately, the crisis gave rise to two overlapping conflictsthe conflict with domestic elites and the conflict with European elites, which both combined economic and political aspects. The double economic and political crises led to a strong alignment of resistance against austerity and calls for democratic renewal (Hutter et al. 2018). The paradigmatic case is Greece, where the imposition of far-reaching austerity programs by the 'Troika' led to the break-down of the ruling social-democratic party (PASOK) and to the rise of a strong new left challenger (Syriza) (e.g. Vasilopoulou 2018).
SE had always looked to the EU as a modernizing force that helped it overcome the legacy of its authoritarian past (e.g. Díez Medrano 2003 for the case of Spain). In other words, Euroscepticism had been weak in SE before the Eurocrisis. As a reaction to the European interventions in their economic policies, we expect, however, a growing resistance against Europe in economic terms. Moreover, we expect this resistance to come mainly from the radical left and less from the radical right which is above all driven by identitarian motives and which has been stronglydiscredited in SE by the previous authoritarian regimes. Thus, the overall impact of the Eurocrisis on southern European party systems is not only expected to be more profound, but, because of the different driving forces also to be quite different from its impact in NWE. Finally, we expect the impact of the Eurocrisis to be complemented by the refugee crisis as Greece and Italy have been very affected by the arrival of refugees across the Mediterranean (e.g. Biermann et al. 2019: 256). As a result, we not only expect an increasing politicization of the immigration issue, but also an increasing politicization of European integration. This politicization is expected to come from the radical right, however, in line with the new structuring conflict in NWE. 1 Finally, the absence of clear-cut cleavages has characterized the political conflict structure in CEE. It has been argued that the Communist inheritance left a fragmented society and an unstructured pattern of political conflict. If measured against the four criteria of institutionalization introduced by Mainwaring and Scully (1995), the party systems in CEE still appear to be poorly institutionalized. They have not (yet) developed stable roots in society, are hardly considered legitimate by the citizens of their countries, their organizations tend to be unstable and they are characterized by extraordinarily high volatility (e.g. Powell and Tucker 2014). Coman's (2017) study suggests that the main dimension of conflict in CEE countries is, indeed, strongly connected to cultural issues. The common denominator of the cultural issues mobilizing the conservative side of the CEE electorates seems to be a defensive nationalism asserting itself against internal (such as ethnic minorities, Roma, and Jews) and external enemies (such as foreign corporations colonizing the national economy).
The expected impact of the Great Recession on the role of European integration for the CEE party spaces is less straightforward than in the other two regions. On the one hand, these countries belong to the main beneficiaries of the integration process, both in economic and political terms. Moreover, the countries in our sample were not members of the Eurozone (i.e. Hungary, Poland, and Romania) or even joined it in spite of the Eurocrisis (i.e. Latvia in 2014). On the other hand, the rise of cultural-identitarian conflicts and defensive nationalism may also be fueled by the integration process. Strategically acting party leaders from the right are likely to mobilize against the EU in the name of defensive nationalism, as is exemplified by Orbán's Fidesz in Hungary (Enyedi 2005) and Kaczynski's Law and Justice (PiS) in Poland. In CEE as in NWE, the nationalist dynamic is likely to have been enhanced during the refugee crisis, fueled by the intergovernmental conflicts in which the governments of some of the respective countries got involved. Nonetheless, given the considerably less structured nature of party competition in CEE and the contradictory political incentives, we do not expect a strong impact of the recent crises on the politicization of Europe by inter-partisan competition in CEE.
Overall, we expect regional and crisis-specific varieties of politicization which reflect the more general restructuration of domestic conflict structures (differentiated politicization hypothesis). That is, we expect conflicts over European integration to be embedded in a reinterpreted cultural divide in NWE, politicized by challengers from the right and particularly since the onset of the refugee crisis. By contrast, in SE, we expect a much more profound effect on the politicization of Europe in the Eurocrisis, driven by challengers from the left and embedded in general struggles over economic and democratic reforms. Finally, we expect the least changes and lowest level of politicization in CEE.

Design and methods
We share Schmidt's (2019) observation that the increasingly politically charged nature of European politics has its source primarily in national politics. As Schmidt also notes, such politicization has engendered what we could call national 'politics against (EU) policyor even politics against (EU) polity'. To study this process, we think it is essential to start with dynamics in national political arenas. Specifically, we analyse debates during national election campaignsas heightened moments of domestic conflictin 15 countries. Six countries represent NWE (Austria, France, Germany, Ireland, the Netherlands, Switzerland, and the UK), and four each SE (Greece, Italy, Portugal and Spain) and CEE (Hungary, Latvia, Poland, and Romania). Given its particular party system and the crises experiences, we consider Ireland together with the hard-hit countries from Southern Europe. 2 For each country, we include in the analysis at least one election before the onset of the Great Recession in the fall of 2008 and all the elections up to the end of 2017. All in all, we cover 58 elections (Appendix A). 3 As argued before, we consider the regional groupings as a helpful heuristic tool but provide empirical evidence on how far they carry us.
We follow our previous strategy and make use of a relational content analysis of newspaper articles to study politicization (Hutter et al. 2016). While massmediated communication is not the only way to study politicization, we consider it a kind of 'master arena' to observe statements in the public sphere. The analysis is based on the coding of two newspapers per country (Appendix A). We selected articles that report on the campaign and national party politics in general during the two months preceding Election Day. We then coded a sample of articles using core sentence analysis. That means each grammatical sentence is reduced to its most basic 'core sentence(s)' structure, which contain(s) only the subject, the object, and the direction of the relationship between the two. For the following analysis, we rely on all coded relations between party-affiliated actors as subject and any political issue as object. The analysis is based on around 73,800 such actor-issue statements.
A crucial step is aggregating the detailed issues that were coded into a set of broader categories. Ultimately, we grouped them into 17 categories (Appendix A). We distinguish between two types of European integration issues. We label them as 'Europe' (which covers general orientations and statements about the further widening and deepening of Europe, except the Eurozone) and 'Euro' (which covers statements about the single European currency, its reforms, exit of some Eurozone members, and bailout conditionalities). The issues were recoded so that positive directions indicate support for European integration and the Euro (including support for Eurozone membership and the bailout agreements).

Empirical results
The data analysis proceeds in three steps. First, we discuss politicization at the systemic level. Second, we map the partisan divides over integration. Finally, we focus on the overall configurations of the political spaces before and during the Great Recession.

Systemic politicization of European integration
For this study, we measure politicization as the multiplication of salience and polarization. By doing so, we treat the two dimensions equally. 4 Systemic salience is measured by the share of core sentences related to a given issue in percent of all statements. The indicator for the polarization of positions is based on Taylor and Herman's index of left-right polarization, and it ranges from 0 to 1 (Appendix A). Figure 1 presents the development of the systemic salience, polarization, and politicization of European integration (including the 'Euro' category) by election. For the interpretation, we added two types of benchmarks. First, we show the corresponding values for immigration as another critical issue related to the new structuring conflict. Second, the two horizontal dashed lines indicate the mean (lower line) and the mean plus one standard deviation (upper line) of politicization calculated for all 17 issue categories. Note that issues that cross the upper line are usually among the top-3 issues in a campaign.
The first graph in Figure 1 highlights that European integration was not a very salient issue in any of the campaigns in the early 2000s. The trend line is below the 'lower' benchmark and the values in any campaign come close to the 'upper' one. However, the salience of Europe has increased thereafter, especially since the beginning of the Eurocrisis in 2010. Also, its level stabilized at a rather high level until the end of our research period in 2017. Remarkably, European integration became more salient than immigrationits 'twin issue' (Hoeglinger 2016: 59)during the Eurocrisis, while the two have become equally (and fairly) salient during the refugee crisis. What is more, European integration became highly polarized since 2010. As a matter of fact, it became one of the most polarized issues (as indicated by the trend line and the many campaigns above the 'upper' benchmark in the second graph in Figure 1). Again, the comparison with immigration is instructive: conflicts over Europe were less polarized in the early 2000s, but they have caught up with immigration by 2012 and show an almost identical upward trend thereafter. The last graph in Figure 1 shows our summary measure of politicization (salience x polarization). It mirrors the trend for salience but shows an even stronger increase during the Eurocrisis given the upsurge in polarization.
Regarding single campaigns, Figure 1 already indicates the extraordinary nature of the four Greek elections in 2012 and 2015. The two types of European issues -Europe and Eurohave become extremely salient in Greece during the Eurocrisis. One can almost talk about 'referendum-like' campaigns in which one issue is at stake only. In combination with the high polarization of the Greek parties' positions, we observe extremely high politicization levels in those four campaigns. This confirms the results of Katsanidou and Otjes (2016) who, based on an analysis of the 2012 elections, highlight how European issues have completely restructured the Greek political space. At the same time, our findings qualify this. As Figure 1 highlights, we see other campaigns in which European issues were very politicized (and at least often highly polarized) since the Great Recession. However, none reaches the extraordinary levels of the Greek cases. The only other Southern European election in which European integration was also particularly politicized was the 2011 Portuguese election (note that this election took place in the shadow of the Portuguese bail-out).
To put these single cases into perspective, more detailed results are presented in Appendix B which show trend lines by region and average values by country and time period. The pre-crisis levels confirm the expectation that European integration was hardly politicized in SE and CEE before the crisis hit. In line with previous research (Hutter and Grande 2014), the overall values for the six countries in NWE in the 2000s are slightly higher but moderate at best. The country-specific results in Appendix B indicate again that the Eurocrisis saw a boost in the politicization of Europe in the national electoral arena in SE. Obviously, the extremely steep trend line for this region is driven by the Greek campaigns. However, the country averages indicate a substantial increase for Portugal, Italy, and Ireland (another hard-hit country), too. The only exception is Spain where European integration became more polarized but was not emphasized much, indicating also the effects of the timing of national elections and party strategies in accounting for within-region variation. For NWE, the trend indicates a substantial increase in politicization since the beginning of the refugee crisis only. This particularly holds for Austria, France, and the UK (of course, the British 2017 campaign was dominated by debates about 'Brexit'). The Dutch 2012 campaign is the only campaign in a 'debtor country' where Europe was not only polarized but also salient during the Eurocrisis. By contrast, Germany and Switzerland saw the most politicized struggles about Europe in campaigns before 2008. 5 Finally, it is important to point to the flat trend line for the countries in CEE. At least during the Eurocrisis, Europe has not become politicized in these countries, and the covered cases during the refugee crisis, i.e. the Polish 2015 campaign and the 2016 Romanian one, do not (yet) deviate from this pattern.

Partisan divides over European integration
We now turn to the level of parties to identify the actors which arebased on our study of public debatesmost strongly associated with opposition to or support for European integration. To do so, we follow Hobolt andde Vries (2015: 1169) and combine a party's EU salience score with the distinctiveness of its EU position (which can be either distinctively anti-or pro-European). 6 Based on this indicator, which is the party-level equivalent of our systemic politicization measure, we identify the parties which shape public debates with salient and distinct EU positions, and reconstruct the changing patterns of opposition in the party system since the onset of the Great Recession. This 'politicizing party' measure constitutes the dependent variable in a series of OLS regression models with robust standard errors. We cross-checked the robustness of our results because of the panel structure of the dataset and potential outliers (for the detailed results, see Appendix C).
To assess the new cleavage hypothesis, we are interested in differences across party families. We grouped the parties into four summary types: radical left 7 , center left (including greens, social democrats and social liberals), center right (including Christian democrats, conservatives, conservative liberals), and radical right. We test the competing strategic hypothesis with two variables: a dummy that indicates whether a party has been in government during the past legislature and the vote share of a party. In addition, we control for the systemic context provided by the other parties as it critically shapes the ability of an individual party to politicize an issue in public debates. Green-Pedersen and Mortensen (2010) have introduced the notion of the systemic agenda to account for the fact that parties cannot just compete on the issues of their choice, but they must relate to issues that are prominent on the party system agenda as well. We adapt their measure to our 'politicizing party' concept: the systemic context is obtained by the product of the average position and the salience of European issues in the party system (always excluding the party in question). We expect a negative relation. That is, a party is more likely to politicize an issue if the other parties in the system emphasize the issue as well but adopt a markedly different position (Meijers 2017).
The first thing that results from the regression analysis is that the systemic context, indeed, has a significant effect on parties' politicization of European issues (for the regression tables and robustness checks, see Appendix C). As expected the effect is negative: we find the strongest Eurosceptic politicizing parties in contexts when the other parties in the system also emphasize the issue but in a more pro-European direction. We find this effect in all three regions. Second, the results confirm the pattern of the inverted U-curve that we expected based on the cleavage hypothesis: the parties from the radical left and radical right are much more likely to politicize in Eurosceptic terms than are the center-left parties (the reference category). This applies overall and in both NWE and SE, but not in CEE. In general, the partisan structuring of European integration is much less pronounced in CEE, which corresponds to the general lack of structure of the party systems and the low systemic levels of politicization described above. The strategic hypothesis also receives some support, although much less than the cleavage hypothesis. However, note that the two 'strategic' effects mainly hold for SE (government vs. opposition) and NWE (larger vs. smaller parties).
To look at changes in times of crisis, Figure 2 presents the marginal effects of 'politicizing party' in a three-way interaction of region, party type, and period. In line with the shifts on the systemic level, we observe the most pronounced changes for the South of Europe. While the two radical party types have already emphasized more Eurosceptic positions before the onset of the Great Recession, we observe a much more distinctive divide between them and the center-left and center-right thereafter (see also Braun et al. 2019). 8 Importantly, it is not just the radical left but also the radical right that is associated with a distinctively Eurosceptic position in SE. We observe the opposite for CEE. In CEE, the distinction into the four party types has even less of a predictive power in the years after 2008. Similarly, the crises have been less of a game changer in NWE. The main change refers to the more Eurosceptic politicizing role attributed to the radical left. A separate analysis of the two subissues suggest that the latter is caused mainly by a higher emphasis of the radical left in NWE on opposition to the Euro. The same holds for the pattern in SE. Mirroring the results for the systemic level and in line with our expectation, the emerging divide in SE iseven more visible in the case of issues related to the Euro and bailout conditionality (for detailed results, see Appendix C).

The configurations of the political space by region
In the last step of the analysis, we focus on the overall configuration of the party systems in the three regions showing how conflicts over European integration are embedded into the political space. Following Kriesi et al. (2008), we construct the configurations with the help of multi-dimensional scaling (MDS). For each region, the respective figure compares the situation before and after the onset of the Great Recession. The regional MDS plots synthesize an enormous amount of information and are the most parsimonious representation of party competition in each region we can conceive. The key strength of the method is that it does not presuppose anything about the dimensions and the configurations of actors. The interpretation of the resulting graphs is up to the beholder. However, the optimal arrangements in our case are two-dimensional (as indicated by the corresponding stress values and scree plots). Moreover, we have rotated the figures in such a way that the horizontal dimension corresponds to the traditional socio-economic divide, ranging from left (a pro-welfare and anti-austerity position) to right (a pro-economic liberalism and pro-austerity position). The vertical dimension corresponds to the cultural dimension whose interpretation, as expected, Figure 2. Marginal effects of the 'politicizing party' score by region, party type, and period. Note: The figure shows the marginal effects of party-level politicization from a three-way interaction of region, party group, and period (pre-2008; 2009-2017). Confidence interval levels 84.4% (i.e. if C.I. do not overlap it means that there is a significant change at α = 0.05).
varies by region. To facilitate the interpretation, we have drawn the horizontal and the vertical axis by joining the issues constituting the polar extremes of the two axes. We have also indicated major clusters of actors by dashed elliptical figures. 9 Figure 3(a) shows the integrated political space for NWE. Overall, the graphs confirm the view of a limited impact of the crises on the structuration of party competition in the region. The space is characterized in both periods by two dimensions: a horizontal economic dimension and a vertical cultural dimension. The latter opposes cultural liberalism and anti-immigration. The issue of 'European integration' is embedded in this dimension as the closeness of 'Europe' to the category 'cultural liberalism' suggests. Within the twodimensional space, we can identify three distinct party clusters in the precrisis period: a unified political left on the upper left and the political right being split between a more centrist and a nationalist-conservative camp. The MDS graph for the crisis period shows an even more unified left camp, on the one side, and a more fragmented political right. The right can now be sub-divided into three distinct clusters. At the bottom, we find a homogenous cluster of all the populist right parties. They are even more closely located to anti-immigration and most opposed to cultural liberalism and Europe, which highlights the populist radical right's predominant focus on this second dimension as does its almost equal distance to welfare and economic liberalism. Figure 3(b) shows the structure of the space in SE. Like the politicization indicators, the graphs indicate a profound transformation of the party systems in SE. The political space does change regarding both the key structuring issues and the main party clusters. In the pre-crisis period, we find essentially the same two dimensions as in NWE: an economic and a new cultural dimension (here the opposition is between cultural liberalism and prodefense because issues of immigration and European integration were hardly salient at all). However, in SE, the economic and cultural conflicts were closely aligned with each other. Reflecting the bipolar type of competition, the two main party clusters in the pre-crisis period consist of a unified political left opposed to the major right-wing parties.
The crisis does not change the dimensional character of the joint Southern European political space. However, what does change is the orientation of the second dimension and its interpretation. It is no longer a 'cultural', but a 'democratic/European' dimension (indicative of the political crisis in the South of Europe) (for country details, see Hutter et al. 2018). Cultural liberalism is much less important in structuring the space as indicated by its peripheral location and the issues 'anti-immigration' and 'defense' are not even represented given their low salience (below 2 percent). 10 By contrast, both European issue categories (Europe and Euro) have become contested. As shown by the party-level analysis, support for both is associated with the center-right and center-left. Also, we find an alignment of conflicts over austerity with those over democratic renewal. Importantly, this alignment is driven by a radical left cluster at the bottom-left of the graph (amongst others, Syriza,  Podemos, and M5S belong to this cluster). This cluster is opposed to a centerleft and a center-right cluster. Thus, while the crisis saw further splits on the right in NWE, it is the left that is split in the South.
Finally, Figure 3(c) presents the results of the MDS procedure for the four CEE countries under scrutiny. The structure of the joint political space comes closer to the one in NWE than SE. We can identify two somewhat independent dimensions: an economic dimension (indicated by the solid line between welfare and economic liberalism) and a cultural dimension (indicated by the line between cultural liberalism and nationalism). The parties are more divided along the cultural than along the economic dimension. In the pre-crisis elections, the second pole of the cultural dimension is associated with positive mentions of nationalism and opposition to ethnic minorities. This contrasts to NWE where it is associated with anti-immigration (given the low salience of immigration, the issue is not represented in the precrisis space for CEE). Thus, as expected, a defensive kind of nationalism, which has been mobilized without the targets of nationalism in NWE, has contributed to the structuring of party competition in CEE. Importantly, the parties in CEE cluster at least as much according to national origins than according to their affiliation with party families.
The space for CEE during the crisis period indicates that the parties are even more differentiated along the cultural divide between cultural liberalism, on the one side, and nationalism, on the other. Note that anti-immigration and Europe are also located in the figure given their increasing salience. Anti-immigration is now associated with nationalism and the conservative right, while Europe is located closer to cultural liberalism (as is democratic renewal). When we look at the location of the different parties in the political space, we see that the differentiation along the cultural dimension is most clear-cut in the cases of Hungary and Poland. The less structured Latvian and Romanian party systems fit less well into the graph and cluster much more regarding national origins.

Conclusions
In this paper, we presented three sets of empirical results about the politicization of European integration since the Great Recession. They allowed us to identify potential changes in the levels of politicization, partisan divides over European integration, and the 'location' of European issues in the regional political spaces.
First, we showed how the politicization of Europe in the national electoral arena developed from the early 2000s to 2017. Our results indicate that the recent crises have acted as 'critical moments' in the politicization process. Overall, the issue of European integration has become much more politicized in the Euro and the refugee crises. However, as expected, we also observe pronounced regional and crisis-specific varieties which reflect the extent of the two crises and the general structuration of party competition. According to our measure, European integration is still hardly politicized in Central-Eastern Europe. By contrast, the Eurocrisis saw a boost in the politicization of Europe in the South and the refugee crisis in the Northwest of Europe. While the Greek campaigns of 2012 and 2015 are exceptional as European integration issues dominated the public debates, there are many other campaigns in Western Europe in which 'Europe' was not just politicized but ranked among the most politicized issues according to our benchmark.
Second, we found the familiar inverted U-curve when analyzing the partisan divides over Europe in public debates, which once again support the 'new cleavage hypothesis'. According to our media data, parties from the radical left and the radical right are more likely to politicize European integration in Eurosceptic terms. The pattern of the inverted U-curve has become more pronounced in times of crises. Again, the change is mainly driven by the countries in the South where this dynamic is linked to significant government vs. opposition divides. By contrast, the results for Northwestern Europe point to stability: only the radical left's opposition to Europe has also become more visible during election campaigns in that region during the Great Recession. Importantly, the partisan structuring of European integration is much less pronounced in CEE and, in the crises years, we cannot discern any significant differences across party types.
Third, we presented the configuration of the party systems in the three regions before and during the crisis to show how the issues of European integration are embedded into the political space. The results suggest a twodimensional space in each region. In Southern Europe, the two dimensions are highly correlated, however, which means that the party systems are more bipolar than the systems in the other parts of Europe. In all three regions, European integration is associated with a reinterpreted second non-economic dimension. However, the association with this dimension varies by region. In Northwestern Europe, European integration marks the opposite pole of its 'twin issue'anti-immigrationand is associated with the left and the liberal forces in the party system. In Southern Europe, it marks the opposite pole to welfare and democratic renewal and is associated with the mainstream forces of the left and right. In other words, the crisis has not changed the structuration of the party system in the Northwest of Europe. If anything, it has served to clarify the structure discerned before the crisis. In Southern Europe, by contrast, European integration was a non-issue before the crisis, and it is only during the crisis that the structure we have just described emerged. In Central-and Eastern Europe, finally, before the crisis the party space resembled that of Northwestern Europe, except that the cultural dimension was not structured by anti-immigration and European integration. The crisis has brought more structure into the party configurations, and it has brought European integration and opposition to immigration into the picture. However, as noted before, the European issues are still much less politicized and, thus, less important in structuring party competition than in Western Europe.
In sum, we take these results as indications that European integration is, indeed, part and parcel of a new structuring conflict that involves a set of processes all of which put the national political community under strain and that restructure national party systems under the impact of the mobilization by the radical left and the radical right. At the same time, our results point to significant regional and crisis-related varieties. Such a pattern of intensified but regionally differentiated politicization constitutes a challenge for both the European Union and recent debates among EU integration theorists (Biermann et al. 2019;Börzel and Risse 2018;Genschel and Jachtenfuchs 2018;Schimmelfennig 2018). On the one hand, the diversity of politicization puts additional stress on a consensus-based political system that is in general not well equipped to absorb and channel political conflicts. On the other hand, our results underline that integration theorists need to incorporate not only questions of 'more or less' politicization, but also its variety when assessing the relative impact of domestic conflicts and Euroscepticism on the future of European integration. Thus, the conclusions about the power of politicization to explain the (different) outcomes of the recent crises have been premature and should be reconsidered in light of our crisis-and regional-specific findings. Notes 1. Note that our research period ends in 2017 and does not include the 2018 Italian election campaign. 2. We cross-checked our results by adding Ireland to the NWE countries. The regional results presented in the paper are not affected by this choice. 3. Given the different timing of national elections, our sample includes no cases since the first peak of the refugee crisis in summer 2015 for Hungary, Italy, and Latvia. 4. We do not consider the third dimension of politicizationactor expansionin our analysis. First, we think that this makes more for sense when studying debates among a wider range of actors (from EU institutions to civil society actors) and for longer periods. By definition, we already observe fairly strong actor expansion if European issues 'make it' into an election campaigns. Second, in this paper, we are most interested in the spatial configuration and the capacity of certain issues to structure that configuration. If an issue is not both salient and polarized, it can hardly structure the partisan space. 5. The Swiss parties fought over the bilateral treaties and the Swiss-EU relations more generally in 2003, whereas the German parties were divided on Turkish EU-membership in 2005. 6. Note that the measure has a direction since the distance from the mean position of the other parties may be negative (in the Eurosceptic direction) or positive (in the pro-European direction). 7. We classified the Italian Five Star Movement (M5S) as radical left. As this is a controversial choice, we ran all models excluding M5S which does not change the results substantively. 8. As the robustness check in Appendix C indicates, the substantive effects (and the corresponding confidence intervals) get smaller if we exclude Greece from the analysis. However, the pattern remains the same: The crisis has reinforced the inverted U-curve in SE. 9. The reader should keep in mind that the MDS-method focuses on the main lines of opposition. Secondary issues/actors are less accurately represented and are often moved to the periphery of the space. Issues that account for less than 2 percent of the observations and parties with less than 30 observations were excluded from the analysis (for a detailed description of the procedure, see Appendix B). 10. While immigration was a high salience issue in the Italian campaign in 2018 (which is not covered by our dataset), the other three Southern European countries showed hardly any significant conflicts over immigration in the campaigns around the first peak of the refugee crisis in 2015.